Molecular Electronics: a Review of Experimental Results
Total Page:16
File Type:pdf, Size:1020Kb
Vol. 115 (2009) ACTA PHYSICA POLONICA A No. 2 Proceedings of the XLII Zakopane School of Physics, Zakopane 2008 Molecular Electronics: A Review of Experimental Results A. Erbe and S. Verleger UniversitÄatKonstanz, FB Physik UniversitÄatsstr.10, D-78457 Konstanz, Germany Molecular electronics aims for scaling down electronics to its ultimate limits by choosing single molecules as the building blocks of active devices. The advantages of this approach are the high reproducibility of molecular synthesis on the nanometer scale, the ability of molecules to form large structures by self-assembly, and the huge versatility of molecular complexes. On the other hand, conventional contacting techniques cannot form contacts on the single molecule scale and imaging techniques nowadays cannot provide a detailed image of such junctions. Therefore, the fabrication has to rely to some degree on self-organization of the constituents. The proof that a molecule has been contacted successfully can only be given by indirect methods, for example by measuring the current transport through the junctions. Here we give an overview of various techniques that were used successfully to contact molecules and to characterize them electrically. The techniques range from methods to contact single molecules to such which can be used to characterize ensembles of molecules. Especially, the comparison between such di®erent techniques shows that a single measurement is always prone to artefacts originating from the unknown microscopic details of the junctions. It is therefore necessary to perform a statistically relevant number of measurements in order to resolve molecular properties. Various properties of the molecules can be studied. Special examples are the influence of conformational changes of the molecules, di®erences between various coupling endgroups of the molecules and e®ects of light-irradiation onto the molecular junctions. PACS numbers: 85.65.+h, 73.63.{b, 85.35.{p 1. Introduction with an electronic system that is divided into two parts, which are separated by a tunnel barrier, could act as The number of transistors in an integrated circuit a rectifying diode. At that time, however, fabrication grows exponentially, approximately doubling every two methods were much too crude to produce structures on years. This behavior was predicted by Moore in 1965 the nanometer scale, which would be needed to contact and is since then known as \Moore's law" [1]. Such an single molecules. It therefore took more than 20 years increase in number of components requires, in turn, to until the ¯rst successful measurements of molecular con- scale down typical component sizes at a comparable rate. ductivity were reported [3{6]. In the years following these It is clear that classical electronic components cannot be initial discoveries, an enormous amount of experimental scaled down without reaching any limits, because quan- results were presented using a large variety of contact- tum mechanical e®ects will dominate the behavior of the ing techniques. An excellent review of the development elements if typical length scales will be of the same or- during these years is given in [7]. On the other hand, der as the Fermi wavelength of electrons (100 nm for it also became clear that reproducible measurements on º typical 2D semiconductors and down to 1 A for metals). the nanometer scale pose extremely di±cult challenges. Gate oxides, which are less than 1{2 nm thick, are also Some of the e®ects which were attributed to molecular no longer reliable due to uncontrollable surface rough- behavior turned out to be artifacts arising from uncon- ness. Due to such reasons, new ways to build electronic trollable variations of the con¯guration on the nanome- devices have to be developed on the nanoscale. One of ter scale [8]. The experience that even the most carefully these ways is the use of single molecules as components performed experiments are prone to such misinterpreta- of electronic circuits. The advantages of this approach tions led to a ¯rst disillusion and the whole ¯eld was said are very clear: The molecular length scale is the smallest to hit an early \mid-life crisis" [9]. The understanding scale which can be imagined for entities carrying electric of these ¯rst experiments made further developments to- current. Molecules can be produced reproducibly in large wards more reliable contacting methods possible. Nowa- numbers by chemical reactions. On the other hand, or- days, there are a number of contacting techniques avail- ganic chemistry can generate a huge number of di®erent able, which have shown reproducible results on various molecules, thus creating a \molecular toolbox", which molecules. In the following, we want to give an overview can be put together for larger scale electronic circuits. on some of these new results. Molecular recognition can then be used to build these circuits in a process called self-assembly. 2. Transport mechanisms The use of single molecules as active electronic devices was already suggested in 1974 in a famous publication by Depending on the molecular structure, the coupling Aviram and Ratner [2]. They proposed that a molecule of the metal to the molecules, and the position of the (455) 456 A. Erbe, S. Verleger Fermi energies of the metals with respect to the molecu- lar energy levels, several transport mechanisms through the molecular backbone can be expected. In the simplest case, no free electronic states on the molecular orbitals are available for the electron coming from the metal, and the electron will tunnel through the distance between the two electrodes. If the electron can occupy some states on the molecule, the transport mechanism will sensi- tively depend on the coupling between the metal and the molecule. If this coupling is good, the coupling be- tween various sites of the molecule will be the de¯ning bottleneck for the transport and activated hopping be- tween these sites will be the transport mechanism in this situation. On the other hand, if the coupling is poor, the metal{molecule junction can be regarded as a tunneling barrier and the molecule will be charged electron by elec- tron. The transport in this regime is generally called the Fig. 1. Tunneling current between two gold tips situ- Coulomb blockade regime. In the following, we would ated a few nm apart from each other. The I¡V charac- like to discuss these transport mechanisms and describe teristics is linear, di®erent resistance values correspond the various I¡V characteristics, which arise from the dif- to di®erent distances. After [34]. ferent regimes. Ã p ! 4d 2m©3=2 2.1. Tunneling I(V ) / V 2 exp ¡ ; (1) 3e~V When no electronic states on the molecule can be ac- where d is the distance between the electrodes, © | the cessed, the electron needs to tunnel through the distance potential barrier, m | the e®ective electron mass and between the two contacting electrodes. In this scenario e | the electron charge. In both regimes the current the height of the tunnel barrier is given by the o®set Ea depends exponentially on the distance between the elec- between the Fermi energy of the metal EF and the closest trodes. molecular orbital, which can be either the highest occu- 2.2. Activated transport pied molecular orbital (HOMO) or the lowest unoccupied molecular orbital (LUMO). Typical energy di®erences are In a molecule that is connected to metal electrodes of the order of 2 eV, for example for gold contacts to sim- with linker groups which provide a good electrical cou- ple alkane chains, where the the HOMO{LUMO gap is pling the metallic and molecular orbitals can hybridize. about 8 eV and the position of EF is determined by the In this scenario the overall conductance of the molecu- work function of gold (5 eV). lar junction is determined by the conductance along the Tunneling is a stochastic process and therefore only molecular backbone. In the easiest case this would be depends on the height and width of the tunneling barrier given by a single molecular orbital, the whole molecule and the rate at which the tunneling particle attempts would simply act as a wire. If the coupling is weaker, to transfer. One of the most important consequences transport is given by hopping of electrons from one metal- is that the conductance of a pure tunneling junction is lic contact to the electronic states of the molecule and to temperature independent. This fact therefore represents the second contact. This hopping is an activated trans- one of the most straightforward tests to verify tunnel- port. The activation can be given either by thermal acti- ing as the dominating transport mechanism. The depen- vation or by an increase in bias voltage. In both cases the dence of the current I on the applied voltage V is deter- transport depends exponentially on V or T , respectively. mined by the amplitude of V . If V is small compared The typical energy scale for these (Eact) is determined by to Ea it determines only the energy di®erence between the coupling between the metallic and molecular states the two electrodes. Therefore I in this regime depends and their respective energies. This transport mechanism linearly on V . This dependence can be veri¯ed for mea- typically results in a so-called s-shaped curve, because the surements of I¡V characteristics between two gold elec- conductance is decreased around zero in a range which is trodes, which are spaced by a few nanometers. The linear comparable to Eact. An example of such a curve is shown dependence is shown in Fig. 1. in Fig. 2b. If the applied bias V exceeds a certain percentage of 2.3. Coulomb blockade the barrier height, the barrier starts to be deformed.